Just tried it out with my webcam, it's pretty laggy; but the potential is huge: www.plugintothesmartgrid.com
Demo: youtube.com/watch?v=NK59Beq0Sew
Just tried it out with my webcam, it's pretty laggy; but the potential is huge: www.plugintothesmartgrid.com
Demo: youtube.com/watch?v=NK59Beq0Sew
yeah started playing with it the other day, really cool, papervision plus cameras == pretty dope.
As I recall - this was the how Snepo did their holiday card. Pretty damn sweet, all in all.
This things get better and better: http://www.youtube.com/watch?v=cTUJKvXIkSU
http://www.bmw.co.uk/z43d
I saw a cool example of this when I was in Portland, but now I can't find it. 
How very clever, sneaking BMW ads into our online counterculture.
=)
Hi! I'm a signature. Wheee!
Originally posted by: Thad
How very clever, sneaking BMW ads into our online counterculture.
=)
Originally posted by: blacksanta_69
It's all fair. If technology is open you have to expect it's going to be used for evil.
And of course - the Japanese have found a way to program this into porn (or nearly).
Probably NSFW - youtube.com/watch?v=yCCx7zANsGE
Originally posted by: rogue_designer
And of course - the Japanese have found a way to program this into porn (or nearly).
Probably NSFW - youtube.com/watch?v=yCCx7zANsGE
really? come on y'all seriously? surely you didn't buy into yet another marketing term? 
Part of a project related email I just recieved:
Silhouette detection with persistent blob primitives of hands and head and silhouette geometry forwarding to Flash would be feasible in that time frame.
The nodes of a very basic skeleton for IK, made up of a tracked head, a torso and two hands could be achieved.
A further option would be add depth map information using 2 cameras in a stereo setup and openCV to allow each silhouette object to have a z value. that way children further away can have different colours.
The data I am sending Flash can be easily formated on my end in either Flash type "array" form or XML form. If you have any other preferences, just let me know.
The information you would receive would be a "Frame ID" with time stamp header for each snapshot of data, and a tracked data object for each person with a unique ID .
Each object will consist of one array of ordered x,y points that make up the outside of the silhouette, and sets of ordered x,y points for the 'holes' in the silhouettes. There should be enough points that will allow for smooth edges without the need for vectors, as noise in vectors can cause exaggerated artifacts.
Also contained within each ID'ed object would be blob coordinates for the head, hands and torso. With this information you can attach primitives to each node at your end.